Title: APPLICATION OF LATTICE GAS AUTOMATA TO PETROLEUM ENGINEERING
1APPLICATION OF LATTICE GAS AUTOMATA TO PETROLEUM
ENGINEERING
- What is it?
- What we are doing.
- What we are going to do.
Assoc. Professor Dr. Mariyamni Awang UTM/PRSS
2PARTICLE MOVEMENTS NAVIER STOKES EQUATION
- Particles moving within a given
structure(lattice) are used to represent fluid
flow. - Reported in mathematical journals before the 80s
but flow in porous media was studied since then. - The first model did not satisfy the NS equation
3First model Four sided lattice, HPP (Hardy,de
Pazzis,Pomeau)
particle
Bigger than a molecule
node
link
4TWO STEPS
- The particles undergo two processes
- Propagation unit mass particle moving at unit
speed in a given direction, along a link
(connection between two nodes). It reaches a
neighbouring node in one time step - Collision occurs under a set of collision rules
and exclusion rules. Only one particle is allowed
to travel in the same direction along a link.
5EXCLUSION RULES
-Maximum of 4 particles to a node at any time,
each one will have a different direction of
movement. - Collision rules result in momentum
and mass being conserved at each node
Againstexc. rule
In
Out
6MODELS THAT WORK
FHP models
Hexagonal Lattice model satisfies NS equation
7FHP MODELS
- 1986 Frisch,Hasslacher and Pomeau used the
hexagonal lattice - Most of the rules for the HPP were maintained and
unlike the HPP, the Navier Stokes equation was
satisfied without introducing any extra term
8MOVEMENT OF PARTICLESFHP 1
- Maximum of six particles to a lattice (for a
hexagon), each particle is of unit mass - Particles collide and move to the neighbouring
node within one time step - The movement takes place along a link and all
particles move at equal speed. - If no collision occurs, movement continues along
the link
9RULES
- Exclusion rules
- Eg Only 1 particle per direction at each node
- 2) Collision rules
- Eg Particles only collide at the node
- 3) Conservation of mass and momentum
10COLLISION RULES
Examples
OR
11FHP II MODEL
- A rest particle is introduced to allow more
flexibility. - Another rule is addednormal particles can
convert to rest particles upon collision
12OTHER MODELS
- FHP III model conserves momentum, allows more
combinations - Face centered hypercube lattice for 3 D
modelling. The collision and exclusion rules are
modified from FHP models
13MORE COLLISION RULES
2 PARTICLES? 1 REST, 1 PARTICLE
14UPSCALING
- - Particles at the microscopic scale, continuum
at the macroscopic scale.
- Macroscopic relationships used are density
,viscosity and pressure to relate to NS Eg
density per node Snumber of particles
- All of the equations have been derived by various
- mathematicians
15LGA Equations, FHP I
16WHY LGA?
- Parallel behaviour
- Outcome of the collisions are not dependent on
the neighbours entirely - No floating point calculations
- Logical ifs only
- New interesting area
17DISADVANTAGES
- Large memory required due to the lattice size,
typical PC allows 300x500 lattices to be
modelled. Node to node distance (O) 1 mm - Incompressible flow only (NV)
- 3D flow has been not been modelled successfully
18REPORTED RESEARCH
- Permeabilities calculated compared with
laboratory measurements were acceptable - Several effects, such as two fluid flow,flow
around obstacles, were reported - Different macroscopic equations were used
19PERMEABILITY
20CURRENT INVESTIGATIONS
- Single phase, 1 D gas flow
- Prediction of permeability for a wide range of
permeability. To be compared with lab results,
core and micromodel. - Problem of representing the physical model on
microscopic level -
Rest particles represent Sand grain
21CURRENT INVESTIGATIONS
- K studies Study the effects of sand grain
distribution, density of fluids, size of
lattice,tortuosity and several boundary
conditions - Two phase displacements lighter particles
displacing heavy particles. -
22FUTURE
-Polymer injection with adsorption effects, dead
end effects. - Cray computer will be used to
increase the size of the model
Application to full scale simulation a la FD
method is not expected in the near future, but
it seems be useful for studies on specific fluid
behaviour and rock characterisation eg
adsorption,capillary flow, diffusion